znnajdla

This happened before with “real” engineering. Engineers a hundred years ago used to be proficient with slide rules and mental math and 3D assembly by hand. Products designed in that era often lasted a hundred years, like Singer sewing machines and Lada cars and 500 year old churches. Now things aren’t built like they used to be, and few people can rebuild civilization by hand like they did. But the world moves on fine!

Read Ben Thompson to understand aggregation theory. Many of the largest internet companies, like Google and Amazon, are simply aggregators. OpenRouter is an aggregator of AI tooling. Stripe itself was just a convenience layer on top of merchant gateways (another aggregator). OpenRouter is so big they can negotiate special contracts with OpenAI for special rates.

Both are the most popular developer api in their space.

Sol is my daily driver but there are still times I reach for Fable when Sol doesn’t cut it. Just yesterday for example, I was trying to build a self-modifying hot-reloaded agent harness in Elixir for fun and Sol just kept doing silly things like thin wrappers and unnecessary abstractions. Fable handled the task elegantly. Sol is really good as a reviewer for finding bugs due to its thoroughness however.

Thanks for putting this so well. The mathematical evidence for this “general intelligence underlying everything” is the “universal geometry of embeddings” paper. Fascinating read. Or as the ancient philosophers used to say, the one who knows God knows everything.

This is a fundamental misunderstanding of how LLMs work. You can’t really specialize a model. You specialize the harness. A well-trained general purpose LLM doesn’t need examples in its training data, it can write good code in a new language you invented yesterday with just a spec definition. And it will perform better than a small model trained on lots of examples of your invented language. The reason is because of the “universal geometry of embeddings”, i.e all human languages have the same underlying pattern structure, so any model that is very good in any language is good in all languages. Attempting to specialize a model for a particular purpose often decreases overall performance. Fine-tuning is just a hack to make dumb models more reliable on limited tasks but they become incapable of doing anything else. Unless you are building a factory assembly line where a model is literally doing the same thing over and over, you almost always want a general purpose model over a specialized one.

I did not claim that eating meat and milk is one's entire point of living. I claimed that the entire point of living is to live according to human nature. Wanting to eat meat and a milk is a part of human nature for pretty much all of humanity's history.

You seem incapable of understanding rhetorical exaggeration.

Is there nothing else going on in your life except to criticize other people's wanting to eat meat and milk, as humans have always done? That I'm sorry for.

Intuitively, the idea makes sense to me. You can only compress something when you reduce the content to “what matters” in it. And understanding “what matters” is to understand the patterns in the data. Understanding the patterns in the data IS intelligence.

There's an important consequence here which I take as a lesson in life and business: it is worth optimizing a process or a workflow in your life or business even when there’s no obvious economic benefit. Because to optimize it is the only way to truly understand it. I am very wary of businesses and software that don’t optimize for performance (not just for profit) because it signals they don’t understand what they are doing. Slow software is poorly understood software. Fast software is also likely to be bug-free and secure because someone understands it.

There's another factor which Ben failed to consider. Which is that NVIDIA doesn't need to rely on demand for their proprietary CUDA stack or their GPUs growing -- they are already selling directly to the consumer, and likely capturing much higher margins. They are moving up stack, not down, where demand for raw compute matters less. With the DGX Spark and Jensen’s statement about “open models”, their next product is likely a strong hint: consumer devices to fulfill the Mac Mini demand craze. They are probably going to start burning LLMs durectly onto sillicon and then selling DeepSeek-in-your-home to individual developers. I bet that would sell even better than Anthropic Max coding plans and is not dependent on hyperscaler funded boom-bust cycles. So Ben’s analysis highlights the risk of their existing business not growing but they are likely planning new businesses.

An indie dev beating Apple at its own game is some feat. This inspires me. Would purchase in a heartbeat.

Very few companies actually have any “trade secrets”. The success of almost any business is based on marketing and distribution and reputation and partnerships, not on any “secret”.

It's one of the classic mistakes of people who are not entrepreneurs. They try to keep things too secret and they spend a lot of time and effort keeping things secret while real entrepreneurs who have actually made millions on the market know secrecy rarely matters.

Very interesting idea but without any concrete examples of performance on real tasks its just a pretty idea

I'm completely new to the AI-security harness space. Can someone tell me what the industry standard is? I just ran Kimi K3 inside Pi to do a quick scan I did find quite a few important things but it seems like it will do better with a specialized harness.

Okay, I get your point.

However, I personally agree with the author and I don't think his argument is necessarily motivated out of an anxious fear. On the contrary I think it may be motivated out of a sense of extreme exhilaration and empowerment.

Because experts (like myself as a programmer for 15+ years) who are using AI in many fields are suddenly empowered and much more useful than we were before AI. My employability and value has gone up and not down, precisely because of being able to apply my expertise with AI, which people without expertise simply cannot do. I am a professional programmer and also owner of my own startup.

Let me give you a concrete example that I am dealing with at my startup. I'm a small business owner. Before AI if i wanted to produce production quality video for marketing it would taken such a huge budget and such a large team of people (or an expensive agency) that I wouldn't even have considered it due to the enormous cost. I'm talking about Apple quality video production which takes millions of dollars to produce.

Not anymore. A single competent person with AI can replace an entire marketing video production department or agency. But expertise is key here: knowledge of film terminology to be able to describe the effect you want, and ability to use video editing tools effectively, as well aesthetic taste. I as a programmer with no filmmaking experience don't even know how to write the prompt which makes the video that i want because I don't even have the terminology. But a person with that expertise has suddenly become more employable and more valuable to my business because I as a small business now have the capability to create Apple quality marketing videos.

So AI actually created a new job for an expert that would have otherwise not existed because it was outside the budget of small businesses. Previously somebody like that would have been employable to only a few large production studios but now they become employable by almost any small business.

What sort of denial are you reading into the author's post? I don't read it as denial. He is fully acknowledging the usefulness of AI.

The conspiracy theorist in me wants to think that the 80% drop in GPT 5.6 Luna prices today is correlated with this update from DeepSeek. Perhaps OpenAI has already hacked its competitors with it's Mythos-like models and is aware of what competitors are doing and is able to react in advance.

One surprising fact is that AI seems to be feeding their business rather than eating it, even though highly technical people have reached a consensus that the quality of the results have gone down. I saw that last night with my non-techie girlfriend who entered a query into the Google search bar which would have been better served by a more powerful model in ChatGPT or Claude. And then I realized that the average person simply doesn’t care or bother to open an app, they still operate on the muscle memory of typing something into Google and expecting an answer. And then the majority of people simply take that answer as fact.

The EU may be right in that the only way to break this hideous monopoly is through fines and regulation, because Google is no longer dominating due to the quality of its service, but on sheer monopoly habits and the muscle memory of a generation of users (and deals with device makers like Apple).

Anybody who is doing well enough to be extremely profitable at their one-person company wouldn't waste time trying to sell $19 guides on how to run a one person company.

Plenty of Americans and Europeans die on the Ukrainian battlefield every day. Western governments don't officially send troops, but there are plenty of people who are willing to risk their lives to defend an oppressed sovereign nation from invasion.

As for the fact that the West could've sent troops but chose not to, yes, that does hurt a lot of Ukrainians, and some do feel used. But we'll fight and win anyway, to defend our homeland, because we have no choice - material support from the West is better than nothing.

This piece of writing moved me to my core, shook my whole being, enlivened my soul. It reminded me that I need to step away from Claude Code every once in a while and read the Classics.

Regarding the Odessa incident— Any government will of course punish traitors. There is no government in the world that doesn’t punish people trying to revolt. The Russian government has killed and imprisoned lots of Russians for opposing the government for example. This Odessa incident is not a proof of anti-Russian hate, it’s a normal response of any government.

Now you agree that Russia started the war to topple Ukraine and and failed in its purpose. So we agree!

I should also add that I still read Dostoyevsky and respect Tolstoy and other high Russian culture. But killing Ukrainians? I don’t respect that. In fact I believe that if Dostoyevsky and Tolstoy were alive today they would oppose the Russian government of today just like they opposed the Russian government of their time. Both were nearly killed and imprisoned by their own government.

> and Crimea and Donbass do not want to be "defended by you".

Most of my friends are from Crimea and the Donbass because my roots are from the Donbass myself. The Ukrainian army is full of people from Crimea (Tatars) and the Donbass who fight willingly to defend Ukraine. Look up the Donbass battalions of the Ukrainian army. Izyum and Kupyansk were cities I visited in 2021 — perfectly safe until Russia invaded and destroyed.

I should also add that I still read Dostoyevsky and respect Tolstoy and other high Russian culture. But killing Ukrainians? I don’t respect that.

In fact I believe that if Dostoyevsky and Tolstoy were alive today they would oppose the Russian government of today just like they opposed the Russian government of their time.

Creeping legitimization of murders and violent actions against people identifying as Russian? Nope I lived in Kharkiv for many years where many people identified as Russian. They felt perfectly safe in Ukraine until Russia invaded. They still feel perfectly safe living in Kyiv and Odessa —- except for Russian missiles of course.

As for Merkel and Hollande — if you think the Minsk agreements or other agreements were in bad faith, then was the invasion of Crimea and Donbass “in bad faith”? There can be no “good faith” agreements when a neighbor chooses to invade you.

> it did not want to start this war

This is such twisted logic it’s insane. If Russia didn’t want to start this war then why did they start it? Who forced them?

I wouldn’t call it “using” I would call it “supporting” since most of us fight willingly. There is no way any army could hold back Russia for 5 years if we weren’t extremely determined and willing to fight.

I’m Ukrainian. I upvoted your comment because I want to show more people how this kind of thinking is mistaken. This kind of logic is very common in those who don’t support Ukraine: they think NATO is “using” Ukraine to fight Russia. They somehow forget that Ukrainians are willingly fighting to defend their own country — NATO can’t “use” a people who are unwilling to fight on their own. (Sure some of my friends have been unwillingly conscripted but that is normal human weakness, that is not evidence that Ukrainians don’t want to defend Ukraine).

If I helped a friend fight off their enemy am I “using” the friend? I saw with my own eyes the thousands of Ukrainians who lined up to fight (and continue to do so) to defend their home country. Sure NATO is supporting us because that’s what allies do, and of course we wish for their increased support. However NATO is not “using” us against our will — they have no ability to compel us to do that.

Sure NATO is already waging war against Russia, but Russia started it.

This. Losing to Ukraine is a massive humiliation. Losing to NATO is “fair” in their eyes. Even now, pro-Russian sources refuse to believe the current drone swarms and highly advanced deep strike drones wrecking havoc on Russian oil infrastructure (see the official state of emergency declared in Crimea and fuel shortages in other regions) are caused by Ukrainian drones. They keep making up some sore loser nonsense about these being NATO drones or some clever trick like drones released from trucks inside Russia. Meanwhile even the Pentagon is buying Ukrainian made drones: https://www.reuters.com/business/aerospace-defense/ukraine-a...

> “I didn’t read it” can’t possibly be an excuse to avoid being bound by an agreement

Only engineers have trouble understanding this. It can be a reasonable defense, and it has successfully been used in courts of law many times. The law is not a machine that compiles text like code literally. Imagine someone who coerces a dying or sick person to sign an agreement they couldn’t possibly be in a reasonable state of mind to understand what they were doing -- the law can and does invalidate such “contracts”. That is the same principle behind age of consent laws. The law could theorerically (and does) invalidate “agreements” which no one is reasonably expected to read and understand.

Sounds like an invasion of Taiwan could fix that problem for them.

Making cheap, “attritable” hardware is fundamentally at odds with raising billions. Cheap drones are winning on the battlefield, but you can’t just spend your way to making something cheap and effective. Effective solutions are born out of grit, will, desperation and sheer necessity, not out of lazy rich greedy spoiled people who’ve never seen a real battlefield investing money in “cheap weapons” because that’s what’s marketable and cool now. And it’s the American taxpayer footing the bill for this immense waste.